Artificial Intelligence and Philosophy Joel Luis Carbonera Gélio José da Silva Júnior Introduction The Interest in the development of intelligent artificial entities participating in the imagination of mankind for a long time. At the same time, attempts to build intelligent robots are also quite old and occurred at various points throughout history, illustrating the intellectual interest in accessing the underlying principles of intelligence. This imagery drew a long way, culminating in the consolidation of a scientific discipline that investigates the issue from the scientific point of view contemporary. Turing's paper in 1950, "Computing Machinery and Intelligence", discusses the possibility of thinking machines, explicitly putting the matter to the relevant scientific community modern and discussing it. This work crystallizes the idea of possibility of a machine behave intelligently, provided that is adequately programmed for this. The name "Artificial Intelligence", and the formal establishment of a research area eat this name occurred in 1956 at a conference at Dartmouth College (Buchanan, 2005). At this conference, research proposals were discussed which orientate the early work of this scientific discipline. Artificial Intelligence Weak and Artificial Intelligence Strong The philosopher John Searle distinguishes two attitudes toward IA: • Strong AI Hypothesis: states that artificial intelligence systems can actually think and have a genuine mind. In this case, these systems would not only be simulated intelligence, but really intelligent entities. • Hypothesis weak Artificial Intelligence: It states that artificial intelligence systems can act wisely, or act as if they were smart (or as if they had minds). In this case, these systems, although they act wisely, would not be genuinely intelligent entities, but in most simulations of intelligent behavior. In this view, computers can be seen as useful tools to study the mind, allowing the empirical testing of models of the mind. Currently, the two views coexist in the area of Artificial Intelligence. A considerable number of researchers in the field adopts the hypothesis of weak AI. On the other hand, there are studies that assume the hypothesis of strong AI, and dedicate themselves to discuss the feasibility of an artificial intelligence general, pursuing the goal of artificial entities that genuinely think, elaborating and proposing theories of how concretely could be done. 1. Nine objections of Alan Turing 1.1: In the article "Computing Machinery and Intelligence", Turing discusses nine possible objections to the possibility of making an artificial intelligence. The teologic objection This type of objection assumes that human beings are endowed with an immortal soul, which is this immortal soul which enables one thought possible and that God has endowed men and women with this strip, but did the same with animals and machines . Thus, neither animals nor machines could think. Turing feels unable to accept any part of this argument, since for some dogmatic views nor women have a soul, which is absurd. Turing also outlines an ironic response, stating that the argument can question the omnipotence of God. If it is God who gives the ability to think, providing a being with an immortal soul, nothing would prevent God to confer with the machines, artificial lighting giving an address for this hypothetical God can fill with a hypothetical immortal soul. Turing points out that, historically, the theological arguments often prove unsatisfactory to the advancement of knowledge and cites the theological objections in relation to the heliocentric theory. 1.2: The “hand in sand” objection Objections of this kind would have the following form "The consequences of machines thinking would be terrible, so we can only hope and believe that this can not be done." Turing sees this argument as popular among intellectuals, proud of their "superior intelligence", who see his position threatened by the possibility of a machine that thinks and what could be more intelligent than themselves. It should be noted that this argument is fallacious because the fact we do not want anything to occur, does not mean it can not occur. The mathematical objection Various results in mathematical logic can be used to show that there are limitations in the discrete state machines (such as modern digital computers). The most well-known results of this type are Gödel's incompleteness theorems, which prove that in any sufficiently powerful logical system, it is possible to construct statements that can not be proved true nor false within the same system. Turing points out that, although it is known that there are limitations on the machines, we can not say that the human intellect has no limitations. Humans make mistakes and it does not shake the confidence that they are intelligent. Even if the human intellect is superior to the machine, it is unclear whether this means that the machine is not smart. 1.3: The argument from conscience In a passage of a prayer, Jefferson said that we could only agree that machine equals brain, when she could write a sonnet or compose a concerto, as the consequences of thoughts and emotions actually felt by the provision and not accidental symbols. That is, to consider an intelligent machine, it would not be enough to write a sonnet, she should also know who wrote it. Jefferson goes on to say that no mechanism could feel pleasure in their successes, or feel pain when its valves fuse, feeling comforted by flattery, feel humiliated by their mistakes, be charmed by sex, be angry or it is depressing when you can not get what you want. Turing see this statement as an assumption of awareness of the need for intelligence. Turing said, according to the argument, the only way we would have to check if a machine is really intelligent, involve being the machine itself and feel yourself thinking. Just as we can but consider another human being as an intelligent being that if we were human. The general position of Turing on this point is that for it is undeniable that there are mysteries to the phenomenon of consciousness, but which, however, is not necessary to solve these mysteries to solve the question of whether machines think. This suggests that Turing see intelligence and consciousness as being independent. 1.4: The inability These arguments imply an inability of the machine, to conclude that, due to this inadequacy, a machine will never be intelligent. Turing raises some of these disabilities, how to be gentle, beautiful, friendly, have initiative, sense of humor, tell right from wrong, make mistakes, fall in love, enjoy strawberries and cream, etc. He points out that, generally, no evidence is offered to support these arguments. Also comments that the reason that people believe these are the disabilities that they had previous experience with limited machinery, special purpose. The Lady Lovelace's objection It is based on reports of Ada Byron (Lady Lovelace) on the analytical engine of Charles Babbage, that the Analytical Engine has no pretensions to originate anything new and what it can do only what we know how to order it to make . However, this does not imply it is not possible build a machine that think for herself, or that, in biological terms, is not possible endow a machine conditioned reflexes that can serve as foundation for learning. The feasibility of this alternative is another question that should be investigated. Turing contrasts this placement of Ada Byron with other issues. How to be sure that an apparently unique and creative work done by a subject was not simply the growth of a seed planted in him by the effect of learning or the fact that it follows the general principles well known? 1.5: The argument of continuity in the nervous system The brain is not a discrete state machine, with characteristics very different from this type of machine. For example, a small difference in signal received by a neuron generates large differences in output signal. Turing was aware of this fact and predicted that it could be used as an argument against the possibility of thinking machines, assuming that these characteristics of the brain thought to be necessary and would not be possible to reproduce these features in discrete state machines. Although Turing agree with the fact that there is this fundamental difference between brains (continuous systems) and digital computers (discrete systems), he disagrees that this fact has consequences that can not a thinking machine. Turing assumes that any system (including continuous systems) can be simulated in a reasonable degree of accuracy in discrete machines, provided they have enough computing power for this . 1.6: The argument from informality of behavior This kind of argument based on the assumption that it is not possible to determine the entire set of rules that describe what a person should do in every possible circumstance, and assumes an additional assumption that there is a set of rules that determines what a machine should do in every situation possible. Turing states that have the behavior governed by a set of rules of behavior implies being a machine, and at the same time and in the opposite way, being a machine implies having the behavior controlled by rules of behavior. Turing notes that it is not possible to say that this set of rules does not exist and therefore we can not say with certainty that human behavior is not controlled by a set of general rules of behavior. The argument of extrasensory perception In 1950 the area of extrasensory perception (telepathy, clairvoyance, precognition, psychokinesis, etc.) was a pretty active area of research. Arguments along these lines would argue that these skills could provide advantages in the Turing test. Supposedly, an interrogator could telepathically identify the man, in a given test. Turing discussed the possibilities to circumvent this. A possible machine sensitive to telepathy, for example, nullify this advantage. Another possibility would be the isolation of the speaker in a room as possible proof of telepathy. 2. John Searle and the Chinese room argument 2.1 John Rogers Searle is an American philosopher who was notorious for his criticism of the projection that the project took IA, mainly in the form of the "Chinese room argument." This argument was presented to the article "Minds, brains and programs" in 1980. This article makes it clear that Searle has no objection to the hypothesis of weak AI, but focuses on outlining a scathing critique of the hypothesis of strong AI. The focus of the criticisms of Searle is the human ability to understand (or understanding). Searle strives to show that computer programs can not display this ability. The philosopher presents a thought experiment that became widely known in the literature as "the Chinese room experiment." • Within a room, there's a human being who speaks English (it could be Portuguese), but does not speak Chinese. • This room has one input channel and output language sentences written • Inside this room, there is a book with rules in English (or Portuguese), which instructs the human being as certain judgments relate to other sentences in Chinese in Chinese. These rules consider only the form of symbols. • It is requested that the subject inside the room, look at the sentences in Chinese entering the room, identify this sentence in the book entry, follow the instructions of the book on this input, generate the corresponding sentence, according to the instructions of the book and put the corresponding sentence in the output of the room. 2.2 In the view of Searle's point of view of an outside observer, the outputs of the room become indistinguishable from the responses of a native Chinese speaker. People who are out of the room can call some of the sentences in Chinese entering the room "questions" and call other sentences in Chinese leaving the room "answers" without the guy who is actually in the room know that comes to questions and answers. Who is inside the room are handling these sentences just comparing the shape of the symbols that come with the symbols of the form provided by the rules of the book, without knowing the meaning of any of that manipulates symbols. With this experiment, Searle wants to show that, even if they are reasonable responses generated in natural language, indistinguishable from a native Chinese speaker would generate, there is no genuine understanding of Chinese in the room, since, just meaningless symbols are manipulated . As a result, even if a machine could pass the Turing test, this does not mean that he truly understands what you are doing. The human ability to understand due to what he called intentionality, a capacity that would be presented by living beings, by which our mental states relate to objects and states of affairs in the world. Searle argues that intentionality can only manifest itself in biological organisms, because it depends on the causal powers of the brain. Searle argues that machines could think if they have causal powers equivalent to the biological brain. 3. Hubert Dreyfus and critique of the assumptions underlying the design of the IA 3.1: Hubert Dreyfus is an American philosopher and in his book "What Computers Can not Do" (1979) he identifies four problematic assumptions that are assumed by researchers in the field, guiding all of AI research. The assumption biological Research at the Beginning of neurology, scientists have assumed that the neuron fired in pulses "All or nothing" (IE, fired or not). This Standard Shooting allowed scientists to see the neuron as logic gates similar to the Digital Computers. This similarity suggested that the brain could be seen as a handler discrete symbols (zero or one). However, Dreyfus shows evidence that a time and the time action shooting neural components have analog components having to do neuron That can not be performed by discrete machines. The assumption psychological From the standpoint of the philosopher, the AI researchers assume, incorrectly, that the mind operates on the information according to formal rules (or at least formalized). For him there is a mass of common sense knowledge which we are unconscious and hard to turn it into an explicit linking of discrete symbols. In view of Dreyfus, this knowledge is not in the brain as a set of individual symbols with meanings, such as AI researchers usually take. 3.2: The assumption epistemological This assumption states that all knowledge can be formalized. This assumption is epistemological, because in Philosophy, Epistemology is the discipline that studies knowledge. The philosopher says that even when the AI researchers agree that psychological assumption is false, may still assume that it is possible that a machine processing of symbols representing all knowledge, regardless of the fact that human beings represent knowledge in this way or not. Dreyfus claims that there is justification for this assumption, since a considerable part of human knowledge would not be symbolic, the point of view him. The assumption ontological This assumption states that the world consists of independent facts that can be represented in separate symbols. This assumption is called ontological, because Ontology is the philosophical discipline that studies the basic categories of things that exist in the world. In view of the philosopher, the AI researchers generally assume that there are limits to formal knowledge, because, like other scientists, assume that any phenomenon of the universe can be described by symbols or scientific theories. Thus, we take everything that exists can be described as objects, object properties, object classes, relationships between objects, etc. Dreyfus casts doubt on this assumption, stating that there is controversy regarding this aspect. 4. Roger Penrose / John Lucas - The mathematical objections The objection is based on mathematical proofs that certain mathematical questions are in principle insoluble to specific formal systems. The Gödel's incompleteness theorems are the best known example of this type of evidence. In summary, these theorems show that, given a formal system (such as a Turing machine), it is possible to make statements within this formal system that can not be proven true or false within the system itself. John Lucas says that Gödel's theorem shows that machines are mentally inferior to humans, because machines are limited by the formal systems incompleteness theorem - can not establish the truth of certain sentences that can express these machines - as humans do not possess such limitation. Penrose says that, unlike the machine, human beings are able to prove such theorems with the use of intuition and creativity and that these phenomena are caused by quantum phenomena that occur in the brain. For Penrose, a machine can not play these quantum phenomena and therefore is not able to think like humans. 5. Philosophical issues • What is intelligence? • How is intelligence? • Machines can think of? • You must have a soul to think? • We must build intelligent machines? • Machines can have free will? • Machines can have emotions? • Machines can be creative? • The brain is a computer? • Used only display the intelligence of its programmers? • The Turing test is crucial to judge the intelligence? • The intelligence requires a body? • Hardware and software are analogous to the brain and mind? • Machines can learn as humans? • Machines can adapt to new situations? • Machines can be conscious? • Awareness is necessary for thought? • Gödel's theorem proves that machines can not think? • Machines can understand natural language? • Machines can make art?
Pages to are hidden for
"ai"Please download to view full document